Study on Risk Assessment and Risk Prevention of Dam Failure During the Operation Period of Tailings Pond

Abstract

There is a huge risk of dam failure during the operation of tailings ponds. Domestic and foreign scholars have conducted extensive research on the assessment and prevention of dam failure risks during the operation of tailings ponds, but there are still many shortcomings. On the basis of exploring the key issues of dam failure risk assessment during the operation of tailings dams, this paper establishes a comprehensive evaluation index system for dam failure risk during the operation of tailings dams based on ten principles including scientificity, systematicity, and operability. By exploring the use of the change statistical mapping method, we can determine the weight of indicators. A risk assessment model was constructed using the fuzzy comprehensive evaluation method; compared to the traditional fuzzy comprehensive evaluation method, this model determines weights in a more extensive and scientific manner. The scientific and effective nature of the model was verified through case analysis of the Shouyun Iron Mine and Shangyu Tailings Reservoir in Beijing. Finally, in response to the risk of dam failure during the operation of tailings ponds, scientific prevention and control measures were proposed from four aspects: personnel risk prevention and control, inherent risk prevention and control of tailings ponds, environmental factor risk prevention and control, and management risk prevention and control.

1. Introduction

Tailings ponds are a major hazard source with high potential energy. Once a dam failure accident occurs, it will cause severe casualties, huge property losses, and irreparable environmental pollution [1,2,3]. According to statistics from 18,401 mines worldwide, the tailings dam failure rate has reached 1.2% in the past century, which is two orders of magnitude higher than the 0.01% failure rate of storage dams. Therefore, accurately evaluating the risk of dam failure during the operation of tailings ponds and implementing effective risk prevention and control measures is of great significance for ensuring the safety of people’s lives and property and protecting the ecological environment [4,5].

In the research of risk assessment index systems for tailings dam failure, scholars at home and abroad have constructed a comprehensive evaluation framework from multiple perspectives. Zhu Yuanle et al. [6] pointed out that the stability analysis and instability mechanism of tailings dams are the core of risk prevention and control, and geological conditions, dam structure, and monitoring technology need to be included in the evaluation system. The risk assessment index system based on correspondence analysis proposed by Salgueiro et al. [7] pays special attention to the special risk factors in the Mediterranean region, reflecting regional adaptability characteristics. In terms of model research, Pérez-López et al. [8] applied machine learning algorithms to risk assessment and developed an improved conceptual Bayesian model that can dynamically learn and update risk parameters. The improved FIM-Unascertained Measurement Model developed by Yangyu Equality [9] significantly improves the objectivity of evaluation by integrating 16 indicators such as flood control intensity and operational stability. Cui Xuyang et al. [10] further introduced dynamic weighted Bayesian networks to enable the indicator system to respond to real-time changes in factors such as rainfall. This type of method better reflects the risk evolution law of the tailings pond operation period by assigning time-varying weights to different indicators. Research by Zheng Xin et al. [11] shows that most models fail to consider the nonlinear effects of fine particle content and saturation of dam materials, while Zhao Haonan et al. [12] have demonstrated through experiments that these factors significantly alter the liquefaction resistance of tailings. In addition, the quantification of management factors such as the completeness of emergency plans and personnel training level in traditional indicator systems is still relatively vague and needs to be optimized in conjunction with fuzzy mathematics theory [13]. Zheng Xin et al. [14] revealed key disaster paths such as infiltration damage and dam slope instability by constructing a fault tree, which was used for identifying the causes of dam failure. Li Xibing et al. [15] used the fault tree method to quantify the structural importance of factors such as rainfall during flood season and the failure of flood discharge facilities. The dynamic weighted Bayesian network model developed by Cui Xuyang et al. [10] successfully captured the risk transition pattern of dam failure probability increasing from 19% to 34.9% under continuous heavy rainfall conditions by embedding time weights. Qu Meixian et al. [16] compiled seventeen important safety risk indices for tailings ponds based on the causes of dam breaches. Dong Yixuan et al. [17], based on safety studies, explored the possible factors leading to dam failure from the perspectives of management, equipment, materials, and environment, and selected these seventeen important risk assessment indicators. On the basis of considering the main causes of accidents in tailings ponds such as overtopping, slope instability, earthquake damage, seepage damage, and management issues, Ke Lihua et al. [18] selected fourteen safety risk indicators for tailings ponds. Galh et al. [19] established a model for the vulnerability threshold curve of victims during landslide disasters by collecting a large amount of factual data and conducting detailed research on hundreds of landslide accidents. Mambretti Galli and his team [20] conducted an in-depth comparison between the laboratory model they constructed and the computational data of the Saint Venant equation and predicted and analyzed the potential impact of dam failure. Hedayati Dezfooli et al. [21] studied the injection molding optimization of propellers using a combination of neural network methods and fuzzy analytic hierarchy process. Lv Zongjie [22] used the Analytic Hierarchy Process to construct a hierarchical structure model for the safe operation of tailings ponds, in order to identify the main factors affecting the safety of tailings ponds. Zhang Minghan et al. [23] selected 10 key indicators to construct a comprehensive evaluation index system for the stability of tailings dam bodies in high-intensity earthquake zones. Ke Lihua and other researchers [24] constructed a tailings dam failure risk assessment model based on the extensible analytic hierarchy process. This model can truly reflect the fuzzy views and preferences of experts on the weight of evaluation indicators for tailings dam failure risk when scoring, in order to improve the accuracy of evaluation indicator weights. Peng Kang et al. [25] used a tailings dam failure risk assessment model and applied the theory of unascertained measures to classify and predict the safety level of tailings dams during operation. They objectively determined the weights of each factor using information entropy theory, effectively eliminating the influence of human factors and making the assessment results more in line with the actual situation. Li Fengjuan and her team [26] proposed a variable weight comprehensive risk assessment technique that combines the Analytic Hierarchy Process and the Entropy Weight Method. This method overcomes, to some extent, the shortcomings of traditional methods, such as strong subjectivity, poor objectivity, and the inability to reflect the relative importance of various influencing factors. Shi Yong and his team [27] used the theory of unascertained measures to explore the relationship between evaluation objects and evaluation indicators. They constructed a comprehensive safety index evaluation system for tailings ponds, including 5 influencing factors and 18 influencing factors, and developed an uncertain measurement model based on an improved entropy weight method.

Research on risk prevention and control mainly focuses on two aspects: technical prevention and control measures, and management prevention and control measures. At the level of technical prevention and control, Wei Zuoan et al. [28] proposed using chemical reinforcement to improve the stability of the dam body, Yin Guangzhi et al. [29] proposed using reinforcement to improve the stability of fine-grained tailings dams, Zhao Yishu et al. [30] proposed using reinforcement strips to improve the anti-sliding stability of the dam body, and conducted research on using other methods [31,32] to improve the stability of the dam body. At the management and prevention level, Liu Mingsheng et al. [33] optimized the design of flood discharge facilities outside the reservoir to avoid cavitation damage inside the structures and reduce damage to flood discharge structures. Prastalo et al. [34] studied the influence of dam failure parameters and evolution curves on the dam failure process line. Fourie [35] proposed applying the Burland soil mechanics triangle theory (topography, soil behavior, and applied mechanics) to tailings dam management to prevent tailings dam accidents from the perspectives of geotechnical engineering and geological engineering. Zhang Yuanyuan [36] proposed corresponding dam failure risk prevention and control methods from the three stages of survey, design, construction, and operation of the tailings pond life cycle, providing strategic support for tailings pond risk management at different stages. Wang Kun [37] put forward improvement suggestions for the prevention and emergency management of tailings dam failure disasters in China in response to the “top pond” issue. K. Stefaniak and M. Wróżyńska [38] took the Zelazny Most (ZeM) tailings pond in Poland as an example and proposed prevention and control measures based on monitoring systems from both technical and environmental monitoring perspectives. There are many similar studies, such as S. Hui et al. [39], J. F. Vanden Berghe et al. [40], etc. There are also applications of new monitoring technologies, such as C. Yaya et al. [41], Sjdahl et al. [42] using resistivity imagers, D. Colombo and B. MacDonald [43] using interferometric synthetic aperture radar (InSAR) technology, and B Schmidt et al. [44] using satellite technology and aerial photography technology, all of which have achieved good results.

In summary, firstly, research on the risk assessment of tailings dam failure can be summarized as qualitative evaluation and quantitative evaluation, or a combination of both. However, most methods still have shortcomings, such as the simplicity of regression analysis algorithms and significant errors. The gray system theory cannot fully consider the complexity of the system. The cloud model theory ignores the fuzziness and randomness of evaluation level information, leading to biased evaluation results. The risk assessment of dam failure during the operation phase of tailings ponds involves numerous influencing factors, some of which can be quantified, while others are difficult to quantify. Therefore, the risk assessment of dam failure during the operation of tailings dams is a typical multi-criteria decision-making problem that requires consideration of the comprehensive effects of multiple indicators. Any purely quantitative evaluation method is inevitably unreasonable, and similarly, any simple qualitative description is also inaccurate. Secondly, research on risk prevention and control either focuses on technology or single-aspect management. Even if there is research on prevention and control throughout the entire lifecycle, it is limited to strategic recommendations and lacks specificity and operability.

The paper adopts a comprehensive evaluation method based on statistical mapping and fuzzy evaluation, which combines qualitative explanation and quantitative description of data investigation to conduct comprehensive analysis and research on multiple factor indicators. It objectively reflects the randomness and fuzziness of tailings pond safety-level information and can comprehensively reflect the opinions of experts on indicator weights and scientifically determine indicator weights. The paper proposes specific prevention and control measures for dam failure risks during the operation phase of tailings ponds, mainly from four aspects: “people, materials, environment, and management”.

2. Key Issues in the Risk Assessment of Tailings Dam Failure During the Operation Period

According to the investigation and analysis of the causes of accidents, the main types of diseases in domestic tailings ponds are shown in

Table 1. Classification and statistics of diseases in domestic tailings ponds.

Disease Classification	Disease Description	Percentage/%
		Black	Other	National	Disaster
		49 Pieces	29 Pieces	78 Pieces	45 Pieces
I	Instability of dam slope	0	3.4	1.3	0
II	Initial dam leakage	8.2	0	5.1	4.5
III	Rainwater causes dam surface collapse	14.3	0	9.0	2.2
IV	Landslide in the reservoir, dam site issues	14.3	13.8	14.1	11.1
V	Pipe surge, soil flow, etc.	20.4	3.4	14.1	4.5
VI	Damage to flood discharge system	32.7	20.8	28.2	33.3
VII	Floods reaching the top	6.1	58.6	25.6	44.4
VIII	Earthquake liquefaction and cracks	4.1	0	2.6	0

[45,46].

2.1. Deformation Problem Under Static Load

The unevenness of materials during the filling process of tailings dams can easily cause uneven deformation, and with the passage of time or the influence of factors such as rainfall, it is easy for the dam body to have many local longitudinal or transverse cracks. In addition, under the action of large flood loads, the dam body undergoes significant settlement deformation, causing floods to overflow the dam crest, directly leading to the failure of the tailings pond and posing a threat to the lives and property of downstream residents.

2.2. Stability Issues of Dam Body Under Static Load

The phenomenon of material creep within the slope, which gradually forms a sliding surface and eventually develops into a landslide, is one of the important factors affecting the safety of tailings ponds. The stability of tailings dam slope is mainly related to several factors: (1) the dam type of tailings dam; (2) the composition and moisture content of tailings; (3) the material characteristics and permeability of artificial dam construction; (4) slope gradient and slope height of the dam; (5) rainstorm.

2.3. Deformation Problem Under Earthquake Load

Under the action of seismic loads, the tailings dam will continue to undergo residual deformation on the basis of static deformation, which may cause more local longitudinal or transverse cracks in the dam body, forming leakage channels.

2.4. Stability Issues of Dam Body Under Earthquake Load

From the situation of several tailings dams in China that have experienced earthquakes, except for the Tianjin Alkali Field, which suffered damage and losses due to liquefaction caused by earthquakes, the other ones mainly showed local liquefaction of the beach surface and individual slopes inside the dam, sandblasting, and water seepage. However, the tailings dam can still be used, indicating that China’s tailings dams have good seismic stability. The experience of earthquake damage to tailings dams at home and abroad shows that tailings dams are prone to liquefaction during earthquakes, causing them to lose stability. Therefore, the analysis of the seismic stability of tailings dams mainly focuses on their ability to resist liquefaction.

Compared with compacted sand shell dams, tailings dams have relatively loose stacking materials, making them more prone to seismic liquefaction. Due to the connection between the dam body and the ore mud in the reservoir, the volume is often much larger than that of an earth dam. In addition, the density of the accumulated material is low, the shear modulus is low, and the natural vibration period of the dam body is quite long. In order to ensure the safety of the calculation results, it is advisable to choose long-period seismic waves transmitted from a long distance when selecting design seismic parameters.

2.5. Flood Control Safety Issues

Flood control safety analysis mainly verifies whether the tailings dam has sufficient safety elevation and dry beach length under the design flood situation, in order to analyze and determine whether the tailings dam will experience flood overflow accidents.

In addressing the aforementioned key scientific issues, attention should be paid to the infiltration line of the dam slope. The seepage line of the dam slope is the lifeline of the tailings dam, and it is one of the important factors directly affecting the safety of the dam body. The analysis of seepage in the dam body runs through all the scientific problems mentioned above, and poor drainage and seismic liquefaction are the main factors causing tailings dam failures. For important tailings pond projects, seepage analysis and dam slope stability analysis under earthquake action should be conducted.

3. Comprehensive Evaluation Index System for Dam Failure Risk During the Operation Period of Tailings Pond

3.1. Principles for Establishing an Indicator System

Establishing an accurate, comprehensive, and effective comprehensive indicator evaluation system is the key to assessing dam failure risk. Based on the analysis and research of the main influencing factors, failure modes, and failure paths of tailings dams during operation, parameters and design indicators related to tailings dam safety can be classified, and a preliminary risk index system for tailings dam failure can be established. To determine each specific influencing factor, it is necessary to extensively solicit expert opinions and refer to relevant regulations for formulation, and then organize, classify, and synthesize them. To establish a comprehensive evaluation index system that can reflect the dam-break risk of tailings ponds during their operation from multiple levels and perspectives, and endow it with certain statistical significance [47,48].

(1)

Principle of scientificity: The selected indicators cannot rely on subjective speculation, but must objectively and truthfully reflect the potential risk of dam failure during the operation of the tailings pond, and measure the development level of each major indicator well.

(2)

The principle of integrating theory with practice: The selection of indicators should achieve a combination of theory and practice. Theory is the guidance of practice, and practice is the extension of theory. Only by guiding practice with theory can practice become more scientific, and only by using theory to serve practice can theory continuously innovate in practice.

(3)

Systematic principle: The established comprehensive evaluation index system should have a clear hierarchy, complete structure, and logical connections between each index.

(4)

The principle of operability emphasizes the feasibility, convenience, and practicality of the selected indicators in practical applications, which directly affects the smooth progress of evaluation work and the accuracy of results.

(5)

Principle of simplicity: The design of evaluation indicators should be as concise and clear as possible, avoiding excessive complexity or redundancy.

(6)

Independence principle: The indicators in the evaluation system should not overlap with each other and can independently reflect different aspects or levels of the evaluated object.

(7)

The principle of comparability: Firstly, choose indicators with relative significance as much as possible, making them easy to quantify. Even qualitative indicators can be compared by establishing evaluation criteria such as excellent, good, medium, poor, and inferior. The second is universality, which means selecting indicators with a wide range of applicability in order to make the evaluation indicators comparable.

(8)

Principle of representativeness: Evaluation indicators should have typical representativeness and accurately reflect a certain characteristic of the risk of dam failure during the operation of tailings ponds. Even with a reduction in the number of indicators, it should be easy to calculate data and improve the reliability of results.

(9)

Principle of comprehensiveness: The selected evaluation indicators should be able to comprehensively reflect the comprehensive characteristics of the risk of dam failure during the operation of the tailings pond.

(10)

Hierarchical principle: The evaluation index system is divided into different levels, each level containing different specific indicators, forming an orderly structure. This hierarchy helps to comprehensively and clearly reflect the different aspects and levels of the evaluation object.

3.2. Indicator System

The risk assessment of tailings dam failure during operation involves numerous influencing factors, some of which can be quantified, while others are difficult to quantify. Therefore, the risk assessment of dam failure during the operation of tailings dams is a typical multi-criteria decision-making problem. Any purely quantitative evaluation method is inevitably unreasonable, and similarly, any simple qualitative description is also inaccurate. The paper adopts a comprehensive evaluation method based on statistical mapping and fuzzy evaluation, which combines qualitative explanation and quantitative description of data survey through multiple factor indicators for comprehensive analysis and research.

Referring to the characteristics of the evaluation object and similar research results at home and abroad [49], the established indicator system is shown in Figure 1, which includes 5 categories of 16 indicators, including overtopping collapse, instability collapse, seepage failure, structural failure, and management factors. This indicator system can provide a prerequisite for the research of risk assessment models and methods for tailings dam failure.

3.3. Determination of Indicator Weights

In the risk assessment of dam failure during the operation of tailings ponds, due to the varying degrees of impact of different evaluation indicators on the risk of dam failure during the operation of tailings ponds, different weights should be assigned to different evaluation indicators to reflect their relative importance. The current weighting methods can be roughly divided into five categories: based on expert consultation values; based on the standard values of indicators as the basis for judgment; based on the measured values of indicators as the basis for judgment; using the measured values and standard values of indicators as the dual criteria for judgment; and other weighting methods.

In order to make the weighting of evaluation indicators more reasonable and the evaluation results more accurate and objective, the method of changing statistical mapping is adopted to determine the weights of indicators. This method is actually a variation of the Delphi method and statistical mapping method. Firstly, list the factors involved in the classification into a table, conduct an expert survey questionnaire, and determine the weights based on the experts’ evaluation of the importance of each indicator. Then, based on the feedback survey information, statistical classification and induction are carried out, and the results are represented by matrix D. When designing a weight order vector, represented in natural number order, C = (n, n − 1, …, 4, 3, 2, 1) (in fact, the difference between the largest, second, …, and smallest weights is represented numerically). Synthesize (map) vector C and matrix D to obtain the weight values of each thematic factor in the comprehensive classification. $\overline{W}=[D\cdot C^T{]}^T$ The weight values are represented in matrix form, which is more comprehensive and scientific than the expert evaluation method. Universality is manifested in the wide range of experts surveyed, and each expert judges the importance of indicators from their own perspective. The scientificity is manifested in the fact that each weight value is calculated based on statistical induction and mapping and can be applied to the comprehensive classification of a large number of thematic factors, and the calculated values can be very accurate.

4. Risk Assessment Model and Case Analysis

4.1. Risk Assessment Model

After determining the weight values of each evaluation indicator, a comprehensive evaluation of the various elements in the risk factors of dam failure can be carried out. Due to the various influencing factors of tailings dam failure during operation, the fuzzy comprehensive evaluation method is adopted here [50,51].

(1)

Establish a factor set, which is the set X of various indicator factors in the risk assessment index system for tailings dam failure during operation.

$$X=\{X_1,X_2,\dots \dots ,X_m\}$$

(1)

Among them, X₁, X₂, …, X_m are m indicator factors participating in the evaluation.

(2)

Establish a weight set, namely the weight set Y of each evaluation indicator factor.

$$Y=\{Y_1,Y_2,\dots \dots ,Y_n\}$$

(2)

Among them, Y₁, Y₂, …, Y_n are the weight values corresponding to the indicator factor X_i, and there are ${\displaystyle \sum _{i=1}^n{Y}_i}=1$.

(3)

Establish an evaluation set, which is a set V composed of the evaluation results of each indicator factor, and then determine the elements in the evaluation set as needed.

$$V=\{V_1,V_2,\dots \dots ,V_n\}$$

(3)

(4)

Establish a single factor evaluation matrix R to represent the fuzzy relationship between the factor set X and the evaluation set V.

$$R=X\cdot V$$

(4)

(5)

Finally, the fuzzy comprehensive evaluation set M is obtained, which involves performing a fuzzy synthesis operation on the weights of each factor and the single factor evaluation.

$$M=Y\cdot R=\{Y_1,Y_2,\dots \dots ,Y_n\}\cdot \left\{r_{\mathrm{ij}}\right\}$$

(5)

Among them, j = 1, 2, …, m.

(6)

The selection of indicator processing methods. Currently, similar studies in China mostly use the maximum membership degree method and the weighted average method for processing. Due to the fact that the maximum membership degree method is prone to information loss, the weighted average method is adopted, which includes:

$$m_j=(y_1{r}_{1\mathrm{j}})+(y_2{r}_{2\mathrm{j}})+\dots \dots +(y_{\mathrm{i}}{r}_{\mathrm{ij}})$$

(6)

where j = 1, 2, …, m.

4.2. Safety Level During the Operation Period of Tailings Pond

According to the basic database data of the tailings pond, as well as relevant specifications such as the “Technical Regulations for Safety of Tailings Pond”, “Engineering Technical Specification for Online Safety Monitoring System of Tailings Pond”, and “Criteria for Hazard Determination of Major Production Safety Accidents in Metal and Non Metal Mines”, the safety level of tailings ponds is divided into four levels, namely A level (normal pond), B level (sick pond), C level (dangerous pond), and D level (critical pond) [13,18].

Referring to the above safety evaluation levels, when applying the fuzzy evaluation theory-based risk assessment model for tailings dam failure, the author suggests dividing the safety level of tailings dams into five levels: Level I (safe), Level II (relatively safe), Level III (sick), Level IV (dangerous), and Level V (critical). Among them, Level I represents that the tailings pond is in a “safe” state and can operate normally, with a safety production license issued. Level II represents that the tailings pond is in a “relatively safe” state. As long as normal management is carried out and the tailings pond does not affect normal operation, a safety production license can be issued. Level III tailings pond belongs to the “hidden danger” status and requires relevant measures to be taken to conduct a hidden danger investigation. It basically meets the conditions for safe operation and can temporarily issue a safety production license. Level IV represents that the tailings pond is in a “relatively dangerous” state, and safety management measures for the tailings pond should be taken seriously. Joint safety agencies should be coordinated to control and manage the pond. If the conditions for safe operation are not met, a safety production license cannot be issued. The Level V warehouse is in a “dangerous” state and has reached the level of imminent accidents. It must be rectified before operation, and if necessary, the warehouse must be shut down. After thorough rectification and acceptance, a safety production license can be issued.

4.3. Actual Case Analysis

The selected tailings pond for evaluation is the Beijing Shouyun Iron Mine and Shangyu Tailings Pond located in Miyun District, Beijing. The bedrock in the reservoir area is ancient gneiss, and the bottom of the ditch is covered by the Quaternary system. The thickest covering layer at the dam site is 16 m. The upper part is mainly composed of alluvial loam, the middle part is mainly composed of slope gravel and soil bearing gravel, and the bottom is a gravel layer. There are two initial dams built in the first phase, with the west dam bottom elevation of 151.6 m and the east dam bottom elevation of 142.5 m. The dam crest width is 5 m, and the inner and outer slope ratios are 1:2. The dam crest elevation is 163.5 m, and the final tailings stacking elevation is 225 m, with a total storage capacity of about 13.7 million m³. The tailings stacking dam is located above the initial dam crest elevation of 163.5 m, and there are five horse tracks on the outer slopes of the dam at elevations of 173.5 m, 183.5 m, 193.5 m, 200 m, and 210 m, with a width of 5 m and a slope ratio of 1:4.5 for each section. In April 2007, the final elevation of the dam body was raised to 245 m, with a total storage capacity of 19.92 million m³. The residential areas and main facilities within 1000 m downstream of the tailings dam include a beneficiation plant, laboratory office building, small repair shop, No. 1 storage, and spare parts storage. In addition, downstream of the tailings dam, there are Jugezhuang Village, Dougezhuang Village, Jugezhuang Railway Station, and about 1000 m to the south, the Beijing Chengde Railway and Mixing Highway pass through, belonging to the “Top Head Reservoir”. The tailings dam storage area is shown in Figure 2 and Figure 3. From the national tailings pond survey system, 16 indicator parameters of the Heshangyu tailings pond can be obtained, as shown in

Table 2. Indicator parameters of the monastic valley tailings pond.

Dam Failure Mode	Indicator	Parameter
Mantan collapse	Flood control standards	500
	Flood discharge facility capacity coefficient	0.8
	The height difference between the beach top and the reservoir water level	1.5
Instability and failure	Average particle size	0.3
	Downstream slope ratio	4
	Current dam height	60
	Earthquake intensity	7
Management factors	Daily management measurement coefficient	0.7
	Accident emergency measurement coefficient	0.8
	Monitoring Equipment Completeness Coefficient	0.9
Seepage failure	Bulk Density	1.7
	Soaking line height	11
Structural failure	Horizontal crack measurement coefficient	0.7
	Vertical crack measurement coefficient	0.8
	Horizontal crack measurement coefficient	0.92
	The integrity coefficient of drainage facilities	0.9

.

The evaluation data was obtained from a questionnaire survey of experts in the expert database. A total of 120 security experts were selected to distribute the survey forms, and 120 were collected, with a recovery rate of 100%. The steps for determining weights using the change statistical mapping method are as follows.

(1)

Organize and statistically analyze the weight information of the survey form, express it in percentage form, and normalize it. Some of the data are shown in Table 3. The weight calculation formula for one-way indicators is:

$$W_{\mathrm{j}}={\displaystyle \frac{T_{\mathrm{j}}}{{\displaystyle \sum _{j=1}^n{T}_{\mathrm{j}}}}}\times \mathrm{100\%}$$

(7)

In the formula, W_j is the weight of a single indicator, T_j is the total score of the importance of the jth one-way indicator given by the expert, and n is the number of one-way indicators.

(2)

Establish a statistical induction matrix D, where C represents the set of indicator importance scores 5, 4, 3, 2, and 1, i.e., C = (5, 4, 3, 2, 1). Use vector C and matrix D to synthesize and map the weight values of each evaluation factor.

$$\begin{array}{cc}\overline{W}& ={\left|D\cdot {\mathrm{C}}^T\right|}^T\hfill \\ & ={\left|\left|\begin{array}{ccccc}0.633& 0.245& 0.136& 0.032& 0.025\\ 0.541& 0.211& 0.134& 0.026& 0.024\\ 0.501& 0.202& 0.121& 0.024& 0.023\\ 0.440& 0.183& 0.105& 0.023& 0.021\end{array}\right|•\left|\begin{array}{l}5\\ 4\\ 3\\ 2\\ 1\end{array}\right|\right|}^T\hfill \\ & =\begin{array}{cccc}(4.642& 4.037& 3.747& 3.319)\end{array}\hfill \end{array}$$

After normalization, $\overline{W}=\begin{array}{cccc}(0.295& 0.256& 0.238& 0.211)\end{array}$.

The weight values of each indicator in Table 3 are shown in Table 4.

Table 3. Summary of investigation and statistics of third-level indicators in part 3.

Index	5	4	3	2	1
Average particle size	0.633	0.245	0.136	0.032	0.025
Downstream slope ratio	0.541	0.211	0.134	0.026	0.024
Current slope height	0.501	0.202	0.121	0.024	0.023
earthquake intensity	0.440	0.182	0.105	0.023	0.021

Table 3. Summary of investigation and statistics of third-level indicators in part 3.

Index	5	4	3	2	1
Average particle size	0.633	0.245	0.136	0.032	0.025
Downstream slope ratio	0.541	0.211	0.134	0.026	0.024
Current slope height	0.501	0.202	0.121	0.024	0.023
earthquake intensity	0.440	0.182	0.105	0.023	0.021

Table 4. Weight values of some tertiary indicators calculated by statistical mapping.

Index	Average Particle Size	Downstream Slope Ratio	Current Dam Height	Earthquake Intensity
weight	0.295	0.256	0.238	0.211

Table 4. Weight values of some tertiary indicators calculated by statistical mapping.

Index	Average Particle Size	Downstream Slope Ratio	Current Dam Height	Earthquake Intensity
weight	0.295	0.256	0.238	0.211

The weight values and ranking of each primary indicator are shown in

Table 5. Weight values and ranking of primary indicators.

Index	Manting Collapse	Instability Collapse	Seepage Failure	Structural Failure	Management Factors
weight	0.128	0.144	0.140	0.148	0.438
sort	5	3	4	2	1

.

From the ranking of the primary indicators in Table 5, it can be seen that the primary factor for the risk of tailings dam failure during operation is still the management factor, and the weight value of the daily management measurement coefficient is the highest among the management factors, indicating that daily management during tailings operation is crucial.

Based on the above evaluation levels $M=Y\cdot R=Y\cdot X\cdot V$, establish an evaluation set V = {Level I library, Level II library, Level III library, Level IV library, Level V library}, and thus obtain a fuzzy evaluation comprehensive evaluation set. The calculation results are shown in

Table 6. Evaluation results of primary indicators.

	Level I	Level II	Level III	Level IV	Level V
Manting collapse	0.408	0.452	0.090	0.043	0.007
Instability and collapse	0.405	0. 456	0.084	0.045	0.010
Seepage failure	0. 394	0.486	0.080	0.034	0.006
Structural failure	0. 401	0. 474	0. 081	0. 036	0. 008
Management factors	0. 394	0. 481	0.085	0.034	0.006

. According to the calculation results in Table 6, the final comprehensive evaluation results of the tailings pond operation period are shown in

Table 7. Comprehensive evaluation results of tailings pond operation period.

Security Level	Level I	Level II	Level III	Level IV	Level V
Comprehensive evaluation results	0.398	0.473	0.084	0.037	0.007

. According to the results in Table 7, 39.8% of experts believe that the Beijing Shouyun Iron Mine and Shangyu Tailings Pond meet the standards of a Class I pond, 47.3% of experts believe that the pond meets the standards of a Class II pond, and 8.4% of experts believe that the pond meets the standards of a Class III pond. From this, it can be concluded that the safety level of the tailings pond is close to the Class II standard of the Class I pond. This evaluation result is consistent with the evaluation of the first-level tailings pond by the China Academy of Work Safety Sciences, which is based on the classification of the tailings pond safety level into four levels.

5. Risk Prevention and Control Measures

During the operation phase of a tailings pond, there are many factors involved in the risk of dam failure, including personnel, processes, management, and the tailings pond system itself. The risk of dam failure during the operational phase appears to be a comprehensive result of various risk factors, but in reality, it is a transmission and evolution of dam failure risks during the survey, design, and construction phases. Regarding the risk of dam failure during the operation phase, prevention and control measures are mainly proposed from four aspects: “people, materials, environment, and management”.

5.1. Personnel Risk Prevention and Control

The prevention and control of personnel risks should first improve the professional cultural literacy of employees. The professional cultural literacy of employees in tailings pond enterprises affects the safe operation of tailings ponds. The professional cultural quality of employees is multifaceted, mainly including four levels of culture: material level (professional facility and equipment operation), behavioral level (standardized job operation behavior), institutional level (sound safety system), and spiritual level (high safety responsibility awareness). Secondly, all types of work in tailings pond enterprises should hold certificates for employment, regularly conduct training on relevant knowledge of tailings pond safety operation and safety production, and regularly review the validity of job certificates. Tailings pond enterprises also need to be equipped with specialized safety management personnel, mainly responsible for managing the safety production of the enterprise, the safe operation of the tailings pond, the behavior safety of enterprise employees, and other aspects. They should also be equipped with dedicated personnel to independently review and manage one or more reservoir archives. In special circumstances, such as subject changes or enterprise bankruptcy, the integrity and traceability of the data can be maintained to ensure the safe operation of the enterprise.

5.2. Inherent Risk Prevention and Control of Tailings Pond

(1)

Control the technical parameters of tailings pond to reduce the risk of dam failure.

The technical parameters such as the height of the tailings dam, the ratio of the dam slope, the safety superelevation, the length of the dry beach, the depth of the infiltration line, and the height of the reservoir water level must be strictly controlled; otherwise, it is easy to cause the risk of dam failure. The height of the tailings dam is the height difference between the top of the tailings dam and the initial dam. The height of the tailings dam should be strictly controlled according to the design. In terms of dam construction technology, upstream dam construction can be used to reasonably control the rising speed of the tailings dam. The slope ratio of the dam during the operation phase mainly refers to the slope ratio of the stacked dam, which is closely related to the slope stability of the stacked dam. It is necessary to determine the slope ratio by calculating the slope stability and considering the impact of earthquakes. Safety superelevation refers to the height difference between the sedimentation beach top of the tailings dam and the design flood level. During operation, if the safety superelevation is insufficient or the length of the dry beach is too short, accidents may occur. The infiltration line is the intersection line between the free water surface formed by the upstream seepage of the tailings dam and the transverse section of the dam body, which is the lifeline of the tailings dam. Its burial depth affects the overall stability of the tailings dam. It is necessary to strengthen the observation and strict control of the burial depth, distribution, and other conditions of the tailings pond without lubrication lines. The high or low water level of a tailings pond affects its safety and stability, and a high water level can also affect its flood control capacity. Therefore, measures must be taken to control the water level of the pond. For example, before the start of the rainy season every year, it is necessary to inspect the flood discharge facilities of the tailings pond to ensure their smooth operation.

(2)

Adopting graded dam construction and composite reinforcement technology to improve the anti-sliding risk capability of dam body.

During the long-term operation of tailings dams, they may experience dam sliding or local instability due to factors such as loading, seepage, and climate change. In response to this risk, it is recommended to adopt a combination of graded dam construction and composite reinforcement to enhance the anti-sliding ability of the dam body and improve structural stability. During the construction or expansion of the dam body, graded stacking technology is used to fill the dam body in sections and layers, strictly controlling the compaction degree of each tailings layer to reduce the impact of uneven settlement on the stability of the dam body. The foundation of the dam body should be subjected to dynamic compaction treatment, or a sand and gravel cushion layer should be used to enhance the bearing capacity of the foundation and reduce the settlement difference in the dam body. In addition, a filter layer can be installed on the surface of the dam slope to reduce the erosion effect of seepage on the dam structure. For tailings ponds that have been built and put into operation, composite reinforcement technology should be used to enhance the strength of the dam body. The upstream side of the dam can be reinforced with an anchor frame, geogrid, or high-strength geotextile to improve the shear strength of the dam slope and reduce the risk of landslides. Internally, cement soil mixing piles or high-pressure jet grouting technology can be used to improve the overall stiffness of the dam body and reduce the possibility of local seepage damage. Under high water level operating conditions, add anti-filter drainage facilities on the upstream side of the dam body to reduce seepage pressure. Further reduce the risk of dam instability [52].

(3)

Building a multi-level drainage system to reduce the risk of infiltration damage.

The tailings pond operates with long-term water storage, and the seepage conditions in the reservoir area are complex. Excessive seepage pressure may cause softening of the dam body, piping, or soil flow, leading to structural instability. To address the risk of seepage damage, it is necessary to construct a multi-level drainage system to achieve efficient drainage of water inside the dam body and reduce the impact of seepage on the stability of the dam body. A three-dimensional seepage control system can be formed inside the dam body by combining vertical drainage wells with horizontal permeable pipes. Vertical drainage wells can be installed on the upstream side of the dam body to lower the water level inside the dam and reduce the accumulation of pore water pressure through infiltration and water collection. Horizontal seepage pipes are installed at the base of the dam body, which can quickly divert the seepage water out of the reservoir area and avoid instability caused by softening of the dam foundation. It is recommended to build interception ditches or anti-seepage walls around the dam body to reduce the impact of groundwater infiltration on the dam body in response to groundwater seepage around the reservoir area. The surface of the dam slope should be drained with open channels or infiltration ditches to divert rainwater and prevent local landslides caused by precipitation accumulation. In the tailings sedimentation area, reasonable drainage ditches should be arranged to quickly discharge surface water from the reservoir area, reduce the moisture content of tailings materials, and improve the consolidation strength of tailings sedimentation bodies. In terms of the operation and maintenance of the drainage system, regularly check the smoothness of the seepage drainage pipes to prevent drainage facilities from being blocked or deformed due to settlement and failure. At the same time, the online water level monitoring system can be used to monitor the changes in seepage water pressure inside the dam in real time, ensuring the long-term efficient operation of the drainage system and reducing the risk of seepage damage [53].

5.3. Environmental Risk Prevention and Control

(1)

Strengthen the inspection of the surrounding environment to reduce disaster risks.

Carry out long-term environmental inspections around the reservoir area, timely control environmental information around the reservoir area, and strictly monitor geological hazards such as landslides and mudslides around the reservoir area. By monitoring the settlement, cracks, leakage, and other conditions of the dam body in real-time, any abnormalities can be promptly addressed (such as thickening the dam body and grouting treatment). Timely maintenance of the dam body, repair of damaged slope protection, erosion prevention of the dam body caused by rainwater flushing, regular inspection of drainage facilities, and regular dredging to avoid blockage and seepage damage are some ways these issues can be addressed and prevented.

(2)

Establish extreme environmental response mechanisms to reduce disaster risks.

Establish a linkage mechanism with the meteorological department, obtain early warning information such as rainstorms and earthquakes in advance, and make emergency preparedness for the risk of tailings pond accidents in extreme environments. Flood control facilities should be optimized in advance to ensure that the flood can be discharged quickly in rainstorms. Flood regulation capacity should be managed in advance to reserve sufficient flood regulation storage capacity. Develop emergency plans in the event of an earthquake, clarify personnel evacuation routes and material allocation plans, reserve emergency supplies such as sandbags, and filter materials to quickly seal dam leakage points. Slope cutting and anchoring of the surrounding mountains in the reservoir area prevent landslides caused by earthquakes from entering the reservoir. At the same time, strictly control the height of the reservoir water level to avoid dam collapse caused by earthquakes during high water level operation.

Through the above measures, the risk of dam failure caused by factors such as dam instability, flood impact, and seepage damage during the operation of tailings ponds can be effectively reduced, ensuring the safety of the reservoir area and the surrounding environment.

5.4. Risk Prevention and Control of Management Factors

(1)

Strengthen management measures to enhance risk prevention and control capabilities.

Management measures are also an important component of risk prevention and control in tailings ponds. Improving management systems and processes is the foundation for ensuring the safe operation of tailings ponds. Tailings pond enterprises should establish detailed tailings pond management systems and operating procedures, clarify the responsibilities and authorities of management personnel at all levels, ensure standardized management of tailings ponds, and strengthen daily inspections and maintenance as key measures to reduce tailings pond risks. Tailings pond enterprises should establish a regular inspection system to conduct regular inspections and maintenance of key parts such as dam bodies, drainage systems, and monitoring and warning systems. By promptly identifying and addressing potential issues, the normal operation of the tailings pond is ensured. The formulation and exercise of emergency plans are also important aspects of risk prevention and control in tailings ponds. In the tailings pond of the metal mine, the mining enterprise has developed detailed emergency plans, including emergency plans for dam break accidents, environmental pollution accidents, etc., and regularly organizes emergency drills for employees to improve their emergency response capabilities and cooperation level.

(2)

Building an intelligent risk warning management system to enhance risk prevention and control capabilities.

The risk of tailings pond has the characteristics of suddenness and concealment, and traditional monitoring methods are difficult to meet real-time warning needs. In this regard, GNSS high-precision displacement monitoring technology should be used for dam deformation monitoring to obtain real-time data on dam displacement and settlement and accurately identify the trend of dam instability. The monitoring of water level and seepage in the reservoir area can be achieved through the use of a fiber optic water level sensing system to automatically monitor water level changes, and combined with AI algorithms to analyze abnormal water level trends and provide early warning of possible seepage damage risks. Acceleration sensors should be installed for earthquake and blasting vibration monitoring to monitor the dynamic response characteristics of tailings ponds under earthquake or blasting conditions, ensuring that the dam remains in a safe state under extreme loads. The intelligent warning system should combine big data analysis technology to conduct trend analysis on long-term monitoring data, identify risk factors, and establish a tailings pond safety assessment model based on historical data. Through the risk level classification and grading response mechanism, when the monitoring data exceeds the set threshold, the system automatically triggers a warning signal and pushes it to the tailings pond management personnel for timely intervention measures. In addition, the early warning system needs to be linked with the emergency management platform to achieve automated risk assessment, accident simulation analysis, and rapid response, improving the intelligence level of tailings pond accident prevention and control.

6. Conclusions

The paper summarizes the current research status of dam failure risk assessment of tailings ponds at home and abroad, explores the key issues of dam failure risk assessment during the operation period of tailings ponds, establishes an evaluation index system and evaluation model, and proposes prevention and control measures.

(1)

Five key issues for evaluating the risk of dam failure during the operation period of tailings dams have been proposed, namely the deformation and stability of tailings dams under static loads, the deformation and stability of tailings dams under seismic loads, and flood safety issues.

(2)

To accurately evaluate the risk of dam failure, a comprehensive evaluation index system for dam failure risk during the operation period of tailings ponds has been established. This system follows ten principles of scientificity and the connection between theory and practice, covering five categories: overtopping collapse, instability collapse, seepage damage, structural damage, and management factors. It uses the change statistical mapping method to determine the weight of indicators, ensuring the objectivity and accuracy of the evaluation.

(3)

On the basis of establishing evaluation indicators, a risk assessment model was constructed using the fuzzy comprehensive evaluation method, and the effectiveness of the model was verified through case analysis of the Shouyun Iron Mine and Shangyu Tailings Pond in Beijing. The safety level of the tailings pond was determined to be Class II, which is close to the standard of Class I, and it was clarified that management factors are the primary factor in dam failure risk.

(4)

In response to the risk of dam failure during operation, targeted prevention and control measures for dam failure during tailings pond operation have been proposed from four aspects: personnel risk, inherent risk of tailings pond, environmental factor risk, and management risk.

In summary, this study systematically completed the risk assessment and prevention of dam failure during the operation of tailings ponds. The established evaluation system and proposed prevention and control measures have important theoretical and practical value. Future research can further explore the application of more scientific methods in risk assessment, continuously optimize risk prevention and control strategies, in order to better ensure the safe operation of tailings ponds and the safety of the surrounding environment, for example, building an AI-empowered comprehensive monitoring system, or establishing an emergency decision support system to transform the risk assessment of tailings dams from “passive response” to “active prevention and control” in the future.